Goal-oriented Energy-aware Multimodal Scheduling for Cooperative Perception

Trong Duy Tran, Maxime Ferreira Da Costa, Salah Eddine Elayoubi, and Linh Trung Nguyen

Abstract

Cellular-vehicle-to-everything (C-V2X) enables cooperative sensing by allowing information exchange among entities such as vehicles, road infrastructure, and cellular networks. A primary challenge in C-V2X research is to achieve energy and communication efficiency. This paper proposed a goal-oriented energy-aware multimodal scheduling scheme for cooperative perception in C-V2X systems. The scheduling policy adapts the activation patterns of sensing modalities, such as cameras, LiDAR, and radar, to maximize sensing classification accuracy. An optimization problem is formulated using a theoretical discriminant gain as a proxy for sensing performance, with constraints on power consumption and communication. An efficient scheme is derived to obtain the optimal scheduling policy. Comprehensive experiments on collaborative V2X datasets demonstrate the efficiency of the proposed scheme, especially under adverse weather conditions.

Cite

@misc{tran:hal-05669525,
  author     = {Tran, Trong Duy and Ferreira Da Costa, Maxime and Elayoubi, Salah Eddine and Nguyen, Linh Trung},
  title      = {Goal-oriented Energy-aware Multimodal Scheduling for Cooperative Perception},
  year       = {2026},
  eprint     = {hal-05669525},
  eprinttype = {hal}
}